Manual assembly needs to be able to react fast and flexible on product modifications at an increasing number of product variants. Cognitive assistance systems and smart devices can increase the efficiency and quality of this product assembly. However, both need to be validated for shop floor application in order to secure the operator’s acceptance. Moreover, available assistance systems force operators to a predefined assembly path and do not support dynamic deviations. Hence, the present paper focuses on the evaluation of smart devices for assembly workstations and the development of a cognitive assistance system for dynamic repair processes. In a comparative study, a smartwatch, a wristband with gesture control and a voice control are compared with an IR remote control and evaluated using defined criteria. Furthermore, different smart devices are connected to an assistance system, which reacts flexibly to a specific error and acts situationally to deviations using a precedence graph.